Hi Harshita,

Thanks for the update. The community review load is currently quite high,
but we will try to review it as soon as possible.

Best regards,
Yufei

On Tue, Oct 27, 2026 at 2:04 PM Harshita Joshi <[email protected]>
wrote:

> Hi all,
>
> I have two PRs that are close to merging and would appreciate a review
> when someone has time. Both are rebased on main, CI is green, and all
> review feedback is addressed.
>
> 1. #5625: Honor catalog table-default properties for staged table
> creates (fixes #4062)
>    https://github.com/apache/polaris/pull/5625
>
>    Staged creates (stage-create: true, used by clients such as DuckDB)
> built the table metadata from the request properties only. Tables
> created that way ended up without the catalog's table-default.* and
> table-override.* properties, while direct creates got them. The PR
> applies them with the same precedence as a direct create.
>
>    flyingImer has approved it. jbonofre, all four of your requested
> changes are in (the LocalIcebergCatalog change is dropped so #5634 is
> untouched, the commit-path test is kept, and the CHANGELOG uses your
> wording). Could you take another look so the "changes requested"
> status clears?
>
> 2. #5869: Reuse AWS service identity credentials across requests for
> SigV4 federation (fixes #4711)
>    https://github.com/apache/polaris/pull/5869
>
>    Each federated SigV4 request built a new
> DefaultCredentialsProvider. Under web identity, for example IRSA on
> EKS, that meant a new AssumeRoleWithWebIdentity call on every request,
> and these providers were never closed. The PR caches one provider per
> realm in an application-scoped cache and closes the cached providers
> on shutdown. There are no SPI or interface changes.
>
>    vigneshio has approved it. It needs a committer review.
>
> Also ready if anyone has time: #5222, a bounded retry for notification
> UPDATEs on concurrent entity modification (fixes #4658). It addresses
> all of flyingImer's earlier feedback.
> https://github.com/apache/polaris/pull/5222
>
> Thanks,
> Harshita
>

Reply via email to